Review questions

1. How do the concerns with consistency differ between NRT and CRT approaches?

2. What does the expression X=T+E mean?

3. Why is correlation important in NRT reliability but less important in CRT dependability?

4. How does test-retest reliability differ from equivalent forms reliability?

5. What are the different internal consistency estimates of reliability?

6. What are the key differences between threshold loss methods, squared-error loss methods and domain score methods?

7. How can you find threshold loss dependability from a single adminstration?

8. What is the IRT assumption regarding unidimensionality?

9. What method is described in this chapter for investigating the unidimensionality of a test?